Register in translated text production

Adriana Silvina Pagano Laboratory for Experimentation in Translation Federal University of Minas Gerais

Register and Context 2011

Context
Project Text production in translation with a view to computational and machine translation modelling: (de)metaphorization in the translation process of expert translators
Federal University of Minas Gerais Universitt des Saarlandes Probral 292/08 CAPES, Brazil/DAAD, Germany

Register and Context 2011

Overview
Background Point of departure #1 translation as a particular mode of text production
Evidence from product Evidence from process

Point of departure #2 register in translated text production

Register complexity in translation Translation as a register/text type

Modeling contextual pressure in translation tasks issues & questions

Register and Context 2011

Background

Register and Context 2011

Translation in SFL
The nearest we get to non-text in actual life, leaving aside the works of those poets and prose writers who deliberately set out to create non-text, is probably in the speech of young children and in bad translations. (Halliday & Hasan, 1976)

Register and Context 2011

Translation in SFL
... translator is very clearly backgrounded in the lexicogrammatical construction of translation - a striking confirmation of Venuti's (1995) notion of the invisibility of the translator. (Matthiessen, 2001)

Register and Context 2011

Translation in SFL
A modelling of translation will force the systemic linguistic community more than hitherto - and give them the tools - to face the interesting challenge of language typology and comparison generally. (Steiner, 2004)

Register and Context 2011

Translation in SFL
... to explain why anyone says anything one must appeal to the context which exerts pressure on the speakers choice of meaning. (Hasan, 2009)

Register and Context 2011

Point of departure # 1
Translation as a particular mode of multilingual text production

Register and Context 2011

Definitions of translation
diachronic production of text within comparable contexts in

different languages, drawing on resources located towards the monolingual pole of the language contact cline [...] translation as the in-context version of typology (Figueredo, 2011:340). multi-functional paraphrase [...] under the constraints of the process of understanding and of the typology of the language systems involved, each individual translation being text production under the constraints of a source text (Steiner, 2004)

Register and Context 2011

A Translation Studies perspective

Evidence from translation product i.e. translated texts as compiled in a corpus and compared to their originals (parallel corpora) and to non-translated texts in the target language (comparable corpora) Evidence from translation process i.e. logs of real-time text production within experimental setting and recorded through keylogging, screenlogging and eyetracking

Register and Context 2011

Universals of translation hypothesis

When compared to non-translated texts in the target language, translations manifest frequencies of linguistic features which systematically deviate from non-translations regardless of the language pair: normalization (standardization), simplification, explicitation and levelling-out (conservatism) (Baker 1993,1995, 1996; Laviosa-Braithwaite 1996, 1997)

Register and Context 2011

Universals of translation
Evidence of explicitation
Higher incidence of optional that in reporting clauses in translated vs. non-translated English (Burnett, 1999; Olohan and Baker 2000) Higher frequencies of reformulation markers such as in other words, namely and that is to say in translated vs. nontranslated English (Mutesayire, 2005)

Evidence of Simplification
Lower type-token ratio in translations (Laviosa-Braithwaite, 1996)

Evidence of standardisation
More restricted use of contracted forms in translated English (Olohan 2003)

Register and Context 2011

Unique items hypothesis

Translated texts would manifest lower frequencies of linguistic elements that lack linguistic counterparts in the source languages such that these could also be used as translation equivalents (Tirkonnen-Condit, 2002)

Register and Context 2011

Unique items
In Finnish (and Swedish), sufficiency lexicalized in verbs jaksaa (Sw. orka); mahtua (Sw. rymmas); ehtiaN (Sw.hinna); riittaNaN (Sw. raNcka); viitsiaN (Sw. idas). In English, no single lexical verb construal as is strong enough; is small enough; is early enough / is quick enough / has enough time; is abundant or sizeable enough; has enough initiative, respectively. When translating from English into Finnish, translators use these lexical verbs less frequently as they do not tend to suggest themselves readily, certainly not as one-to-one equivalents to any particular item in the source text.
Register and Context 2011

Literal translation automaton hypothesis

Tendency to translate word by word observed in novices as well as experts, and in process as well as product data. Literal translation as a default unless interrupted by a monitor that alerts about literal renditions that are not linguistically acceptable or contextually appropriate as translation equivalents. Same evidence found in experiments with bilinguals (Mandelblit, 1996) sentences with different mapping in source and target languages take longer to translate and are first attempted through literal rendition (Tirkonnen-Condit, 2005)

Register and Context 2011

SFL-informed translation studies

Steiner (2001, 2004), Teich (2003), Neumann (2008) Combined use of parallel and comparable corpora Cross-register analysis SFL- informed analysis

Register and Context 2011

Shining Through and Normalization

Teich (2003)
English originals and English translations from German have a higher frequency of passives than German translations from English and German originals English translations have more passives than English originals TL normalization (due to the tendency for translations to exaggerate the typical features or patterns of the TL) There are more passives in the German translations than in the German originals SL shining through (due to interference from the SL)
Register and Context 2011

Explicitation
Neumann (2008)
When contrasted with a reference corpus, translations in the FICTION and LETTERS TO THE SHAREHOLDERS registers, unlike originals in these two registers, show an increase in the number of conjunctions - explicitation in terms of conjunction in these registers in the translation direction English-German

Register and Context 2011

Properties of translated texts

Steiner, 2001
Three sources to account for properties of translated texts typology of source language system reflected in some of the properties of the translation (literal translation) registers of source text and target text for a given context may not be the same - translator(s) may decide on changes oriented to the register of target-text understanding on translators part involves relating given units of text to more explicit and more literal paraphrases

Register and Context 2011

(De)metaphorisation hypothesis
Understanding in monolingual and multilingual text production can be modelled based on grammatical metaphor as relating meaningful (grammatical) units to their less metaphorical variants = explicitation of meanings in original text drawing on co-textual and contextual knowledge. De-metaphorization taking place at same time as re-wording in target language poses constraints on translators work due to typological and/or registerial reasons, or fatigue

Register and Context 2011

(De)metaphorisation
Translation from English into German (more nominal-type phrases translated into more verbal-type phrases)
The suspicion that volcanic eruptions are the primary source of aerosols in the upper atmosphere has been around for many years.
(English Original)

Seit vielen Jahren vermutet man schon, dass die Aerosole in den hheren Schichten der Atmosphre vor allem aus Vulkanausbruchen stammen. (Translation)
Since many years suspects one already that the aerosols in the higher layers of the atmosphere above all stem from volcanic eruptions (word-for-word backtranslation into English)
Register and Context 2011

Shifts in levels of metaphoricity

Metaphorisation - TL expression more metaphorical than SL expression De-metaphorisation - SL expression more metaphorical than TL expression Re-metaphorisation - same degree of metaphoricity in SL and TL expressions (implies de-metaphorisation and can only be observed in translation process!)
Source text - Im Jahr 1982 kommt Howard Schultz zu Starbucks. Target text 1 - The year of 1982 brings Howard Schulz to Starbucks. Target text 2 - In 1982 Howard Schultz joins Starbucks.
Register and Context 2011

In search for evidence in the process

In the case of demetaphorization and metaphorization in the product, are these solutions arrived at in first renditions or worked out in several renditions? In the case of re-metaphorization, is there evidence of de-metaphorization as a previous step in the process?

Register and Context 2011

Translation process
Alves, Pagano, Neumann, Steiner & HansenSchirra, 2010
Keylogging records the history of every group and clause in text production, from first renditions, interim solutions and final choices as they can be seen in translation product - Pauses are taken as indicators of effortful production Screenlogging records computer activity (web/dictionary/online databank searches) Eyetracking records areas of interest and foci and movements of fixation Recall protocols record subjects account for his/her decisions
Register and Context 2011

Sample ocurrence
Source text Wir sind davon berzeugt, dass erfolgreiche Unternehmensfhrung und soziale Verantwortung sich nicht widersprechen.
We are of that convinced that successful management and social reponsability do not contradict each other

Target text We are convinced that successful management and social responsibility are not contradictory.

Register and Context 2011

Aligned words with different PoS tags

verb

TT1: We firmly believe that success business and social responsibility do not necessarily contradict each other.
verb

ST: Wir sind davon berzeugt, dass erfolgreiche Unternehmensfhrung und soziale Verantwortung sich nicht widersprechen.

TT2: We are convinced that successful management and social responsibility are not contradictory.
Instance of grammatical shift. Explanation?
adjective!

Rank shift: Phase noun Verb

TT1

TT2
Verb!

Original Drafting

sich nicht widersprechen

are not contradictions in do not contradict Rank shift: terms

Verb noun

Drafting
Drafting Revision Revision

do not necessarily contradict each other

Back to verb; effect: no change in metaphoricity

do not contradiction
are no contradiction are not in conflict are not contradictory
Rank shift: Noun adjective; effect: change in metaphoricity

Registerial constraints
Evidence from recall protocol:
3:34: mixture of grammatical and lexical problems wasnt sure whether I wanted to use a nominal or a verbal construction thats why I came back to this later 14:06: I mentioned that before I was in conflict (laughter) I didnt like conflict because it seemed too (?) for a corporate text they wouldnt use negative words thats why I changed it back to contradictory I think

Implications
Two behavioral patterns specific of translators (Toury, 1995) bearing an impact on translated text production
1. Relying on literal translation as a first strategy for rendition 2. (Un)packing more or less metaphorical wording in an effort to understand the source text or due to typological and registerial constraints

Register and Context 2011

Point of departure # 2
Register as a relevant notion to model translation

Register and Context 2011

Register in Translation Studies

Long acknowledged as a key concept as the translators proceeds in his/her text production, he/she calibrates register Attempts to explore register in translator education either lack an encompassing theory of language (Newmark 1988, Nord 1991, Stolze 1992, Wilss 1982) or, when they do draw on a theory, do not work with the notion of register in translation at a level of delicacy to tackle its complexity (cf. House 1977/97, 2001, Bell 1991, Hatim and Mason 1990, 1997, Taylor 1998). Within SFL studies, register in language contacts (multilingual production, translation) has not been sufficiently explored

Register and Context 2011

Register complexity in translation

If the values for register variables are similar in two cultures, translation, unlike multilingual text production, entails a relative stability of register (Steiner, 2004) BUT Changes introduced in the target text in one registerial dimension may have impact on other dimensions of variation of a text Register variables may change as a text unfolds, which adds more complexity to the translators task.

Register and Context 2011

Register complexity in translation

If systems of register features (variables) and their values differ between languages and the values for register variables are different, the translator may need to create a register or to introduce a change in the target text in one registerial dimension, which may have impact on other dimensions of variation of a text.

Register and Context 2011

Register complexity in translation

Metaphorisation as a strategy to deal with contextual constraints We hope that the Australian Embassy services in Brazil will be able to accommodate this short time frame (original)

... esperamos que a concesso do visto por parte das autoridades da Embaixada Australiana no Brasil no se veja prejudicada pela solicitao sem a devida antecedncia (translation)
... we hope that the granting of the visa by the authorities at the Australian Embassy in Brazil will not be affected by the request filed without the due amount of time in advance (back translation)
Register and Context 2011

Register complexity in translation

There may be explicit instructions in the situation in which translation process takes place to change register values e.g. changes in tenor There are social norms regarding translation: what a translation is expected to look like, literal wording being a valid strategy for some texts or even yielding a valid target text, though not a felicitous one from a comparable corpus perspective (see also Juliane Houses distinction between overt and covert translation)

Register and Context 2011

Literal strategy
Bilingual edition of popular tales for foreign language learners
Hace mucho tiempo vivi un molinero que tuvo tres hijos, y nada ms que su molino, su burrito y un gato para ellos. Long long time ago, there lived a Miller with three sons and nothing else than his mill, his little ass and a cat for them.

Register and Context 2011

Register (in)stability and translation

Impact from another language system excerted through registers deliberatly mimicking source language registers
Register (in)stability due to (in)stability of target text register Register (in)stability due to (in)stability brought about by the act of translation

Register and Context 2011

Register (in)stability and translation

Tourism promotional texts cross languages Cross-cultural Register Studies Group
Adriana Pagano, Universidade Federal Minas Gerais, Brazil Ayako Ochi, Macquarie University, Australia Maria Herke, Macquarie University, Australia Marvin Lam, City University of Hong Kong, Hong Kong Mira Kim, New South Wales University, Australia Pattama Patpong, Mahidol University, Thailand Sabine Bartsch, Technische Universitt Darmstadt, Germany Stella Neumann, Aaachen University, Germany,
Register and Context 2011

Feature matrix for the analysis

Style Indicators
Content orientation Addressee orientation

Casual

Consultative

Neutral

No

Yes (?)

Yes

Yes

Yes

No

Shared background

Yes

No

No

Register (in)stability and translation

Tourism promotional texts in Argentinian Spanish
Changes in the system of social distance in Argentinian Spanish increasing use of form of address vos for interactions previously calling for usted Tourism promotional texts in Spanish modelled upon texts in English need for addressing the reader (you) Random and sometimes concomitant use of vos and usted in non-translated texts Norm dictates that translated texts cannot be too different from originals (size and structure)

Register and Context 2011

Register complexity in translation

Source text markedness (Steiner, 2004) (no conformance to the expectations of readers) may demand conscious decisions on the part of the translator Translators agenda may be to draw on literal translation to introduce foreignization ( Venuti 1995)

Register and Context 2011

Translating the register(s) in fiction

How to deal with register(s) in fiction that are based on real life registers but construe a particular style?
Katherine Mansfields Bliss in Portuguese and Spanish
Should the translator base his/her choices on features of more or less colloquial registers in his/her language so as to re-construe the style of Bliss? Should the translator build on the target language literary system and construe a style patterned on practitioners of Mansfields style in those target systems? Would Brazilian writer Clarice Lispector be a model for a translation of Mansfield to be patterned on? Which female writer could be a model for a translation of Mansfields style in Spanish?
Register and Context 2011

Bliss
ALTHOUGH Bertha Young was thirty she still had moments like this when she wanted to run instead of walk, to take dancing steps on and off the pavement, to bowl a hoop, to throw something up in the air and catch it again, or to stand still and laugh at - nothing at nothing, simply. What can you do if you are thirty and, turning the corner of your own street, you are overcome, suddenly by a feeling of bliss - absolute bliss! - as though you'd suddenly swallowed a bright piece of that late afternoon sun and it burned in your bosom, sending out a little shower of sparks into every particle, into every finger and toe? Oh, is there no way you can express it without being "drunk and disorderly"? How idiotic civilisation is! Why be given a body if you have to keep it shut up in a case like a rare, rare fiddle? "No, that about the fiddle is not quite what I mean," she thought, running up the steps and feeling in her bag for the key - she'd forgotten it, as usual - and rattling the letter-box. "It's not what I mean, because - Thank you, Mary" - she went into the hall. "Is nurse back? "Yes, M'm."

Register and Context 2011

The register of translations

BUT even in situations in which no deliberate decision is made to foreignize
and no deliberate instruction is given to follow a literal strategy or to change register parameters,

due to source text constraints demanding typological and registerial

alignments and constraints related to translators understanding ((de)metaphorization) and reliance on lexicogrammatical wording of the source text (literal strategy), translated texts have properties that make them different from other texts. In this sense, translations can be seen as a register / text-type in themselves (Steiner, 2001, 2004)

Register and Context 2011

Register in translation
Questions
How to accommodate the register of translations within a SFL account of register? If translation is a register, how to describe its contextual features? What is the impact on translator education and training? Would expansion of learners registerial repertoires (Matthiessen, 2009) include the register of translation?
And many other...
Register and Context 2011